MVA- DRAFT

NOT FOR DISSEMINATION

2017-2019 Residential Property Sales

Median sale ($)

Recoded for any value > $250,000

Median sale/sqft ($)

Recoded for any value > $150/sqft

Properties sold at least once (%)

Cluster analysis

The method for classifying Cleveland’s neighborhoods into a typology is a two step process. First, neighborhood features relevant for understanding the local housing market are entered into a principal component analysis (PCA). PCA works by taking several pieces of information, or variables, about a neighborhood that can be directly observed and measured (e.g., home sales, housing density, etc.), and using the underlying relationships between those variables to uncover new variables that cannot be directly observed or measured, but that may be important defining features of the neighborhood. The variables entered into the PCA were:

• Median sales price per square foot
• Variation in local sales prices
• Number of foreclosures
• Percent of residential properties with recent sales
• Rental housing units per residential structure
• Percent vacant residential structures
• Percent multi-family properties with Section 8 contracts
• Percent of land area zoned commercial property class
• Number of demolitions

The PCA distilled the relationships between these nine variables into two new components:

• Component 1: an index of distress that describes low home values, above average demolitions and foreclosures, and relatively low sales rates

• Component 2: an index of density, defined by above average rates of rental housing and commercial property area

The components were then used as inputs into a cluster analysis. The clustering algorithm settled on a solution categorizing block groups into five neighborhood types/clusters.

Neighborhood Types by Median Sale/sqft

Distribution of sale prices by cluster

Cluster Descriptives